Application Of Machine Vision Recognition Technology In Picking Robot System

Machine vision recognition technology is the key to the transformation from traditional production to intelligent production.In the system of the picking robot,the visual software receive the image from image acquisition equipment,the information including position and rotation,is extracted according to the pixel distribution and the gray value in the image,process this information and send to the lower computer which drive the robot to pick.After the integration of machine vision,picking robots work more efficiently,stability and safely,which can replace and surpass munual picking.This topic aims to develop the visual software in the picking robot system,researching model making,image matching,calibration and application.In the section of model making,design a median filter for image smothing filtering,enhance the contrast of the image contour,make the image contour more prominent.In the contour extraction method,a contour extraction method based on Canny algorithm is designed.Then research the operation principle of Halcon operator,debug its internal parameters through a large number of experimental comparision,complete model making.In the section of image matching,analyzed several common image matching method,after comparing their advantages and disadvantages,choose the shape-based matching method.Subsequently,an improved image matching process is reseached,based on the Hausdorff idstance matching algorithm with pyramid structrue,design matching solution,match the target object in the image.The optimization strategy adopted in the conflict between the matching speed and the thoroughness is proposed,and the matching process with the high speed and stability is realized greatest degree.In the section of calibration and application,the principle of software calibration is described in detail,and an optimized software calibration method is designed.The calibration of the camera internal and external parameters is realized by using the operator in the function library.According to the structure of the picking robot,the result of matching is converted into the data that can be directly executed by the lower computer,and the communication of the upper and lower computer is realized through the EPA message.Finally,in the.NET framework using Halcon library to write the software.The experimental results show that every functional modules can work normally,the performance meets the robot system requirements,the robot can realize intelligent picking with the participation of visual software.